Core Viewpoint - SoftBank, under the leadership of Masayoshi Son, is strategically positioning itself to become the world's leading provider of Artificial Super Intelligence (ASI) by investing heavily across the AI and semiconductor value chain, from IP to application layers [5][10][37]. Group 1: Historical Context and Vision - Masayoshi Son's journey began in 1975 when he was inspired by a microcomputer chip photo, which ignited his lifelong commitment to technology and innovation [6][9]. - In the 2025 fiscal year report, Son articulated a new strategic goal for SoftBank: to become the foremost ASI platform provider, emphasizing the belief in the eventual emergence of intelligence surpassing human capabilities [9][10]. Group 2: Strategic Investments - SoftBank has made significant investments in various companies to build a comprehensive AI and semiconductor ecosystem, including a $20 billion investment in Intel, becoming one of its top shareholders [13]. - The Stargate project, in collaboration with OpenAI and Oracle, aims to construct large-scale data centers for AI infrastructure, with an estimated investment of up to $500 billion [14]. - SoftBank led a $40 billion financing round for OpenAI, indicating its commitment to both infrastructure and application layers in the AI stack [16][19]. - The acquisition of Ampere for $6.5 billion aims to fill gaps in SoftBank's CPU capabilities, enhancing its position in the cloud computing and AI inference markets [20]. - The purchase of Graphcore, a struggling AI chip company, allows SoftBank to diversify its AI accelerator technology portfolio [21]. Group 3: Capital Map and Ecosystem Integration - SoftBank is constructing a capital map that integrates various components of the AI and semiconductor ecosystem, from IP (Arm) to CPUs (Ampere) to AI accelerators (Graphcore) and manufacturing (Intel Foundry) [23]. - The strategy involves creating a closed-loop system that connects upstream IP with downstream applications, thereby enhancing SoftBank's influence in the AI sector [27][28]. Group 4: Arm's Role and Future Prospects - Arm remains a crucial asset for SoftBank, with the company holding approximately 90% of Arm's shares post-IPO, which is pivotal for revenue generation through licensing and royalties [26][30]. - Arm's business model, characterized by long-term benefits from initial licensing, positions it well for sustained revenue growth, particularly in emerging markets like AI and cloud computing [30][31]. - The potential development of proprietary chips by Arm could further solidify its position in the data center market, although it presents challenges and risks [31][32]. Group 5: Competitive Landscape - SoftBank's approach contrasts with Nvidia's vertical integration strategy, as it seeks to leverage capital to control various segments of the AI and semiconductor landscape without focusing solely on in-house development [34][35]. - Unlike cloud giants like Microsoft and Amazon, which emphasize self-developed chips and infrastructure, SoftBank aims to reorganize production factors across the ecosystem, culminating in applications like OpenAI [35][36].

Core Viewpoint - The AI chip market is shifting focus from training to inference, as companies find it increasingly difficult to compete in the training space dominated by Nvidia and others [1][20]. Group 1: Market Dynamics - Nvidia continues to lead the training chip market, while companies like Graphcore, Intel Gaudi, and SambaNova are pivoting towards the more accessible inference market [1][20]. - The training market requires significant capital and resources, making it challenging for new entrants to survive [1][20]. - The shift towards inference is seen as a strategic move to find more scalable and practical applications in AI [1][20]. Group 2: Graphcore's Transition - Graphcore, once a strong competitor to Nvidia, is now focusing on inference as a means of survival after facing challenges in the training market [6][4]. - The company has optimized its Poplar SDK for efficient inference tasks and is targeting sectors like finance and healthcare [6][4]. - Graphcore's previous partnerships, such as with Microsoft, have ended, prompting a need to adapt to the changing market landscape [6][5]. Group 3: Intel Gaudi's Strategy - Intel's Gaudi series, initially aimed at training, is now being integrated into a new AI acceleration product line that emphasizes both training and inference [10][11]. - Gaudi 3 is marketed for its cost-effectiveness and performance in inference tasks, particularly for large language models [10][11]. - Intel is merging its Habana and GPU departments to streamline its AI chip strategy, indicating a shift in focus towards inference [10][11]. Group 4: Groq's Focus on Inference - Groq, originally targeting the training market, has pivoted to provide inference-as-a-service, emphasizing low latency and high throughput [15][12]. - The company has developed an AI inference engine platform that integrates with existing AI ecosystems, aiming to attract industries sensitive to latency [15][12]. - Groq's transition highlights the growing importance of speed and efficiency in the inference market [15][12]. Group 5: SambaNova's Shift - SambaNova has transitioned from a focus on training to offering inference-as-a-service, allowing users to access AI capabilities without complex hardware [19][16]. - The company is targeting sectors with strict compliance needs, such as government and finance, providing tailored AI solutions [19][16]. - This strategic pivot reflects the broader trend of AI chip companies adapting to market demands for efficient inference solutions [19][16]. Group 6: Inference Market Characteristics - Inference tasks are less resource-intensive than training, allowing companies with limited capabilities to compete effectively [21][20]. - The shift to inference is characterized by a focus on cost, deployment, and maintainability, moving away from the previous emphasis on raw computational power [23][20]. - The competitive landscape is evolving, with smaller teams and startups finding opportunities in the inference space [23][20].

在这种格局下，新晋芯片企业在训练市场几乎没有生存空间。"训练芯片的市场不是大多数玩家 的竞技场"，AI基础设施创业者坦言，"光是拿到一张大模型训练订单，就意味着你需要烧掉数千 万美元——而且你未必赢。" 如果您希望可以时常见面，欢迎标星收藏哦~ 在AI芯片这个波澜壮阔的竞技场上，一度被奉为"技术圣杯"的大规模训练，如今正悄然让位于更 低调、但更现实的推理市场。 Nvidia依然在训练芯片市场一骑绝尘，Cerebras则继续孤注一掷地打造超大规模计算平台。但其 他曾在训练芯片上争得面红耳赤的玩家——Graphcore、英特尔Gaudi、SambaNova等——正在 悄悄转向另一个战场：AI推理。 这一趋势，并非偶然。 AI训练作为一个重资本、重算力、重软件生态的产业，Nvidia的CUDA工具链、成熟的GPU生态 与广泛的框架兼容性，使其几乎掌握了训练芯片的全部话语权。而Cerebras虽然另辟蹊径，推出 了超大芯片的训练平台，但仍局限于科研机构和极少数商业化应用场景。 正因如此，那些曾在训练芯片上"正面硬刚"Nvidia的创业公司，开始寻求更容易进入、更能规模 化落地的应用路径。推理芯片，成为最佳选项。 Gr ...